Hailuo AI / MiniMax
Complete Company Profile
A current, evidence-led research dossier covering MiniMax Group and its Hailuo AI creative platform: corporate identity, founders, leadership, timeline, models, Hailuo H3, technology, pricing, funding, IPO, financials, users, partnerships, case studies, API, open-weight strategy, legal/safety issues, competition and strategic lessons.
Information verified as of 16 September 2026.
VERIFIED Primary-source facts are separated from REPORTED independent reporting and INFERENCE analysis.
Supplied images embedded in this standalone report.
1. Executive Summary
2. Company Overview
| Entity | Type | Status | Evidence |
|---|---|---|---|
| MiniMax Group Inc. | Multimodal AI foundation-model company | Public, HKEX 00100 | VERIFIED |
| Hailuo AI | AI-native creative/video platform | Current product | VERIFIED |
| Nanonoble Pte. Ltd. | Singapore legal/service entity for Hailuo services | Current operator named in Terms | VERIFIED |
| MiniMax Open Platform | Developer/enterprise API platform | Current | VERIFIED |
3. MiniMax vs Hailuo AI
MiniMax
- Listed corporate group and foundation-model developer.
- Builds language, video, speech and music models.
- Operates AI-native products and Open Platform.
- Reports company-wide financials and user/developer metrics.
Hailuo AI
- Creative application/platform layer.
- Centered on video/image/multimodal creation.
- Uses MiniMax video and multimodal models.
- Legal service terms name Nanonoble Pte. Ltd. as operator.
4. Founding Story, Mission & Values
MiniMax's public profile says founded in early 2022; corporate materials point to December 2021. Publication-safe wording: “established in December 2021; publicly described as founded in early 2022.”
Mission
Intelligence with Everyone — “与所有人共创智能.”
Values
No Shortcuts · User-in-the-Loop · Tech-Driven
Strategy
The 2025 annual report describes a transition from foundation-model company toward platform company, emphasizing intelligence density, token throughput and global commercialization.
5. Founders & Leadership
| Person | Role | Background | Evidence |
|---|---|---|---|
| Dr. Yan Junjie | Founder, Chairman, Executive Director, CEO, CTO | Former SenseTime VP / vice-head of research; Southeast University mathematics; CAS AI doctorate; Tsinghua postdoctoral research. MiniMax says ~200 papers and 30,000+ citations. | VERIFIED |
| Yun Yeyi | Executive Director, COO; identified as co-founder in current company communications | Former SenseTime fundraising/strategy and innovative business; Johns Hopkins BSc electrical engineering with economics/mathematics. | VERIFIED |
| Zhao Pengyu | Executive Director; LLM research leader | Joined Aug 2023; former Hulu Technology; Peking University BSc/MSc. | VERIFIED |
| Zhou Yucong | Executive Director; visual-model research leader | Joined Mar 2022; former SenseTime/Huawei; Beihang mathematics/computer science. | VERIFIED |
6. Dr. Yan Junjie — Documented Record
Mathematics, Southeast University; AI doctorate, CAS Institute of Automation; postdoctoral research at Tsinghua.
MiniMax IR reports approximately 200 academic articles and more than 30,000 citations.
MiniMax IR lists Wu Wenjun AI awards, Guangdong Technology Invention Award and Shanghai Oriental Talents recognition.
7. Company Timeline
VERIFIED
REPORTED
REPORTED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
VERIFIED
8. Corporate / Product Structure
9. Hailuo AI Overview
Users
Creators, marketers, filmmakers, designers, social teams, e-commerce teams and developers.
Workflow
Prompt/reference → generate → iterate/edit → export; H3 adds multimodal reference and native audio.
Beyond T2V
Camera control, references, editing, audio and agentic workflow orchestration expand the product beyond a basic text-to-video tool.
10. Hailuo Product Evolution
| Model/product | Date | Capabilities | Status |
|---|---|---|---|
| T2V-01-Director | 03 Mar 2025 | Text-to-video camera control | Legacy |
| I2V-01-Director | 03 Mar 2025 | Image-to-video camera control | Legacy |
| Hailuo 02 | 18 Jun 2025 | Native 1080p; instruction following; physics; later start/end frames | Superseded |
| Hailuo Video Agent | 20 Jun 2025 | One-click/template workflow | Evolved into Media Agent |
| Hailuo 2.3 | 28 Oct 2025 | Motion, stylization, micro-expression; T2V/I2V | Legacy relative to H3 |
| Hailuo 2.3 Fast | 28 Oct 2025 | Faster/lower-cost batch creation | Legacy relative to H3 |
| Media Agent | 28 Oct 2025 | Multimodal model/workflow selection and one-click creation | Current direction |
| MiniMax H3 | 31 Jul 2026 | Omni-modal text/image/video/audio; native stereo audio; 4–15s; 24 FPS; 768p base + 2K regeneration | Current flagship |
11. Hailuo 2.3
Released: 28 October 2025. MiniMax reported improvements in physical action, dynamic expression, stylization, micro-expressions and motion-command response. API documentation supports 768p 6/10s and 1080p 6s configurations.
Fast: MiniMax said 2.3 Fast could reduce batch content-creation costs by up to 50%. This is a company-reported model claim, not independently verified customer ROI.
12. MiniMax H3 — Current Flagship Video Model
| H3 specification | Public evidence |
|---|---|
| Modalities | Text, images, video and audio as unified context. |
| Output | Video with native stereo audio; 4–15 seconds in open release; 24 FPS. |
| Resolution | H3-Base 768p; H3-Regenerate-2K provides 2K regeneration using original context. |
| Aspect ratios | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 among documented ratios. |
| References | Up to 9 images, 3 videos and 3 audio clips; 12 mixed files maximum. |
| Dialogue languages | 11 stable languages in the open release, with additional varying support. |
| Architecture | 33B dense H3-Omni-Transformer plus visual/audio VAE components. |
| Encoder | Qwen3-VL-32B used by H3's encoder according to MiniMax documentation. |
| Hosted components | H3-Context-IR and H3-Regenerate-2K are not fully open-sourced in the initial release; official API workflow is provided. |
13. H3 Technical Research in Plain English
Contextual Omni Representation
A hosted context-processing layer parses relationships among text, images, audio and video before generation.
H3-Omni-Transformer
A single dense Transformer jointly predicts video and audio latent representations.
H3-VAE
Visual/audio VAEs compress media into latent representations for efficient generation.
In-context regeneration
The model reuses original context and the 768p output to regenerate a 2K result rather than relying on a conventional standalone super-resolution model.
14. Hailuo Director Models
T2V-01-Director and I2V-01-Director launched globally on 3 March 2025. Their defining purpose was camera control: creators could specify professional-style camera movement and framing. These are legacy models relative to H3.
15. Hailuo Video Agent → Media Agent
Video Agent
Beta launched 20 June 2025. MiniMax described a pipeline spanning brainstorming, scripting, image generation, video, voiceover and editing.
Media Agent
Introduced 28 October 2025. It automatically matches multimodal models/workflows and supports one-click or step-by-step creation.
16. MiniMax Foundation Models & Products
| Product/model | Category | Status | Primary role |
|---|---|---|---|
| MiniMax M3 | LLM/agent | Current | Coding, agents, long context |
| MiniMax M2.7 | LLM/agent | Current | Coding, workplace, agents |
| MiniMax M2.5 | LLM/agent | Prior/current ecosystem | Coding/productivity |
| MiniMax H3 | Video/multimodal | Current | Video + audio + multimodal references/editing |
| Speech 2.8 | Speech | Current | TTS/voice |
| Music 3.0 | Music | Current research/product line | Music generation |
| MiniMax Agent | AI workspace | Current | AI-native work/agents |
| MiniMax Code | AI coding | Current | Coding |
| MiniMax Design | AI design | Current | Design |
| MiniMax Audio | Audio | Current | Speech/music |
| Talkie | Consumer AI companion | Current | Interactive characters |
| Open Platform | Developer/enterprise | Current | APIs, token plans, enterprise services |
17. Technology, Training & Data
Publicly known
- Multimodal R&D across language, video, speech and music.
- Cloud/training costs are material to R&D.
- H3 architecture is unusually detailed publicly.
- H3 includes public weights, inference code and deployment guidance.
Not publicly complete
- Full proprietary training-dataset inventory.
- Complete media licensing/provenance list.
- Exact total training FLOPs/cluster size.
- Independent audit of all training-data provenance.
18. Benchmarks
| Benchmark | Evidence | Limitation |
|---|---|---|
| Artificial Analysis Video Arena — H3 | Current H3 page reports Elo around 1302.42; Hailuo's dated H3 page reports category scores of 1,242 T2V-with-audio, 1,130 video editing with audio and 1,185 I2V-with-audio as of 31 Jul 2026. | Blind human preference; scores change as votes accumulate. |
| MiniMax internal benchmarks | Company reports strong coding/productivity results for M2.5 and later models. | Not equivalent to independent evaluation. |
19. Business Model
AI-native products
Hailuo and other applications generate direct product revenue. H1 2026 AI-native product revenue was $42.6M company-wide.
Open Platform / enterprise
API calls, token plans, inference and enterprise services. H1 2026 revenue was $73.9M.
Developer monetization
Pay-as-you-go APIs, video packages, token plans and enterprise arrangements.
20. Pricing — Current API Snapshot
Pricing verified on 16 September 2026.
| Model/service | Published price | Notes |
|---|---|---|
| MiniMax H3 768P API | $0.08/sec | Current pay-as-you-go. |
| MiniMax H3 2K API | $0.13/sec | Current pay-as-you-go. |
| H3 768P → 2K regeneration | $0.05/sec | Per regenerated output second. |
| Hailuo 2.3 Fast | $0.19 / 768p 6s; $0.32 / 768p 10s; $0.33 / 1080p 6s | Legacy API pricing. |
| Hailuo 2.3 / 02 | $0.28 / 768p 6s; $0.56 / 768p 10s; $0.49 / 1080p 6s | Legacy API pricing. |
21. Funding & IPO
| Period | Event | Evidence |
|---|---|---|
| 2021–22 | Early financing; MiHoYo/Yunqi/Future Capital appear in secondary reporting, exact amounts not consistently disclosed in primary sources. | REPORTED |
| Jun 2023 | >$250M fundraising reported by Reuters. | REPORTED |
| Mar 2024 | At least $600M Alibaba-led financing reported by Bloomberg at valuation above $2.5B. | REPORTED |
| Jan 9 2026 | HKEX listing; final offer price HK$165; stock code 00100. | VERIFIED |
22. Financial Performance
| Metric | 2024 | 2025 | H1 2025 | H1 2026 |
|---|---|---|---|---|
| Revenue | $30.5M | $79.0M | $30.4M | $116.6M |
| Gross profit | $3.7M | $20.1M | $3.7M | $20.8M |
| Gross margin | 12.2% | 25.4% | 12.1% | 17.9% |
| R&D expense | $189.0M | $252.8M | $124.3M | $296.9M |
| Adjusted net loss | $(244.2)M | $(250.9)M | $(138.7)M | $(293.0)M |
2025 annual figures are from the HKEX annual report; H1 2026 figures are unaudited company-reported results. These are MiniMax Group-level, not Hailuo-only.
23. Users & Global Reach
24. Customers & Partnerships
Public evidence is strongest for integrations, creative collaborations, festivals and developer/creator ecosystem relationships. Public named Hailuo enterprise-customer disclosure is comparatively limited.
| Organization | Relationship | Product | Documented activity | Date |
|---|---|---|---|---|
| VEED | Launch/integration partner | Hailuo 2.3 | Hailuo 2.3 in VEED AI Playground | 28 Oct 2025 |
| Envato | Integration partner | Hailuo 02 | Hailuo 02 in VideoGen | 3 Jul 2025 |
| VideoTube | Distribution/integration partner | Hailuo 2.3 | Hailuo 2.3 in AI Video Generator | 14 Nov 2025 |
| Monks | Strategic creative partnership | Hailuo/MiniMax | Creative workflow collaboration in Greater China | 2025 |
| fal | Technology/API ecosystem | Hailuo models | Global creator/developer ecosystem events | 2025–26 |
| SIAFF | Film-festival collaboration | Hailuo + MiniMax Audio/Image | Showcase and AI Toolkit Guidebook | 2025 |
| WSXA Amsterdam / Around Films | Film competition | Hailuo AI | Global AI film competition | 2025 |
| HKU AI Society | Education/community | Hailuo AI | Open Day, masterclass, workshop | Feb 2026 |
| GMO + AiHub Japan | Community ecosystem | Hailuo AI | Tokyo GenAI meetup | Jun 2025 |
| WAIFF | Film/creative partnership | MiniMax Hailuo | Global partner / Chinese special track | Apr 2026 |
| Stellar Pictures | Creative collaboration | Hailuo/multimodal models | The Fleeting Beauty music video | Apr 2026 |
| The Dor Brothers | Creative collaboration | Hailuo | Death Lane showcased at Cannes | Apr 2026 |
25. Case Studies / Creative Collaborations
| Project | Industry | Model | Classification | Workflow/result | ROI |
|---|---|---|---|---|---|
| VEED | Creative SaaS | Hailuo 2.3 | Verified integration | T2V/I2V inside VEED; editing remains in VEED | No public ROI |
| Envato | Creative marketplace | Hailuo 02 | Verified integration | Hailuo 02 added to VideoGen | No public ROI |
| VideoTube | AI video platform | Hailuo 2.3 | Verified integration | T2V/I2V in ready-to-post shorts workflow | No public ROI |
| Monks | Advertising agency | Hailuo/MiniMax | Verified partnership | Early access and workflow collaboration | No public quantified result |
| Stellar Pictures | Film/music video | Hailuo/multimodal | Verified collaboration | The Fleeting Beauty combines AI video, music and Eastern aesthetics | No public ROI |
| The Dor Brothers | AI filmmaking | Hailuo | Verified collaboration | Death Lane showcased with Hailuo at Cannes | No public ROI |
| Antonia Kim / AITONIA | Creator/education | Hailuo AI | Verified creator collaboration | Tutorials, workshops and creator workflow demonstrations | No public ROI |
| Casa Nacho | Food brand demo | Hailuo 2.3 + Media Agent | Demonstration, not customer | MiniMax demoed a one-click 30-second ad workflow | Not applicable |
26. Before → After Workflow
Traditional
- Concept/script
- Actors/assets
- Studio/camera or animation setup
- Voice/audio recording
- Editing/compositing
- Localization/versioning
- Review/export
Hailuo/MiniMax
- Prompt + references
- Generate video/audio
- Iterate via instructions/references
- Use Media Agent where suitable
- Edit/refine
- API-automate at scale where suitable
This is a workflow comparison, not a claim that every customer used the exact traditional process.
27. Before/After Metrics Database
| Company/product | Metric | Before | After | Improvement | Source |
|---|---|---|---|---|---|
| MiniMax | FY2025 revenue | $30.5M (2024) | $79.0M (2025) | +158.9% | HKEX Annual Report 2025 |
| MiniMax | H1 revenue | $30.4M | $116.6M | +283.1% | MiniMax H1 2026 results |
| MiniMax | Open Platform + enterprise AI revenue | $9.2M | $73.9M | +703.1% | MiniMax H1 2026 results |
| MiniMax | AI-native product revenue | $21.2M | $42.6M | +100.9% | MiniMax H1 2026 results |
| MiniMax | Gross profit | $3.7M | $20.8M | +464.8% | MiniMax H1 2026 results |
| MiniMax | Gross margin | 12.1% | 17.9% | +5.8 pp | MiniMax H1 2026 results |
| MiniMax | R&D expense | $124.3M | $296.9M | +138.8% | MiniMax H1 2026 results |
| MiniMax | Users | 236M+ at Dec 2025 | 300M+ current | Company-reported growth | MiniMax official profile |
| MiniMax | Enterprise customers/developers | 214,000 at Dec 2025 | 2M+ current | Company-reported growth | MiniMax official profile |
| Video ecosystem | Cumulative video generations | — | 600M+ by end-2025 | Scale milestone | 2025 Annual Report |
| Hailuo 2.3 Fast | Batch creation cost | 2.3 baseline | Up to 50% lower | Company-reported claim | Hailuo 2.3 release |
| H3 | Video duration | Hailuo 2.3 max 10s | 4–15s open release | Longer maximum | H3 open release |
28. API & Developer Platform
| Capability | Evidence | Use |
|---|---|---|
| Text-to-video | Official video_generation API | Programmatic short-video generation |
| Image-to-video | Official API | Animate product/creative images |
| First/last frame | Official API | Controlled transitions |
| Subject reference | Official API | Character/face consistency |
| H3 multimodal | Official H3/Open Platform docs | Video + audio + references |
| Video Agent/templates | Official API guide | Template-based workflows |
| MCP/CLI | Official GitHub organization | Agent/tool integration |
29. Industry & Use-Case Analysis
Advertising / Creative
Strong public collaboration area: agency workflows, brand demos and creative projects.
Film / Media
Director models, film competitions, Cannes collaborations and AI filmmaking.
E-commerce
Hailuo 2.3 positioning explicitly includes commercial content/e-commerce; product-ad creation is demonstrated.
Education
University workshops and education collaborations are documented; quantified ROI is limited.
Software / SaaS
VEED and Envato demonstrate model distribution inside existing creative workflows.
Creator economy
Global creator programs and product design emphasize rapid short-form creation.
30. International Expansion
MiniMax reported deployment in 200+ countries/regions in 2025 and about 73% of FY2025 revenue outside mainland China.
SIAFF collaboration, creator education and university workshops.
GMO + AiHub Japan community partnership.
Global distribution is visible, but H3 open-weight licensing has territorial exclusions.
31. Competitive Landscape
| Competitor | Overlap | Neutral observation |
|---|---|---|
| Kling AI | Video | Overlapping T2V/I2V and creator workflows; quality varies by task/version. |
| Runway | Video/editing | Strong overlap in professional creative workflows; different product/model ecosystems. |
| Google Veo | Video + audio | Overlapping audiovisual generation and commercial workflows. |
| ByteDance Seedance | Video | Overlapping creator/commercial video production. |
| OpenAI Sora | Generative video | Overlapping creative generation; access/workflows differ. |
| Pika / Luma / Adobe Firefly | Creative generation | Compete for creator/marketing workflows with different features and integrations. |
No universal winner is declared. Benchmark results are task- and date-dependent.
32. SWOT Analysis
Strengths
- Broad multimodal portfolio
- Fast model iteration
- Global reach
- API/enterprise growth
- H3 multimodal + audio
- Developer ecosystem
Weaknesses / gaps
- Large R&D spending/losses
- Limited public named Hailuo enterprise data
- H3 license restrictions
- Rapidly changing access
Opportunities
- API-native commercial video
- Multimodal agents
- Creative SaaS distribution
- Open-weight ecosystem
- Global developer adoption
Threats
- Model commoditization
- Intense global competition
- Copyright/personality rights
- Compute economics
- Regulatory/export controls
33. Copyright, Legal & Responsible AI
H3 license
The Community License excludes the EU, UK, South Korea and US from the general Applicable Territory. Commercial products/services above $20M annual revenue require prior written authorization.
Safety
MiniMax says H3 inputs and enhanced prompts are subject to automated moderation; suspected unlawful, pornographic or infringing content may be blocked. False positives/negatives are acknowledged.
Hailuo Terms
The Aug. 19, 2026 Terms update clarifies responsibility around AI-generated content, authenticity/disclosure and downstream uses.
Training data
A complete itemized training-data inventory is not publicly disclosed in the reviewed sources; no specific dataset is asserted here without evidence.
34. Creator Experience & Negative Evidence
Documented strengths
- Multimodal references
- Native stereo audio
- Camera/motion/voice controls
- API access
- Agentic workflow automation
User-reported limitations
Google Play reviews include complaints about generation quality, credit/cost limits, account/support issues and inconsistent results. These are anecdotal reports, not controlled studies.
35. Open-Weight Strategy
H3 is a strategic shift: MiniMax released weights and inference code publicly. GitHub/Hugging Face provide local deployment resources, while the custom license preserves geographic and commercial controls. The strategy combines openness with managed legal boundaries.
36. Top 20 Technology & Business Milestones
Technology
- Multimodal foundation-model stack
- Hailuo video generation
- Director camera control
- Hailuo 02 native 1080p
- Start/end-frame workflows
- Video Agent
- Hailuo 2.3
- Hailuo 2.3 Fast
- Media Agent
- 600M+ video generations
- H3 native stereo audio
- H3 multimodal references
- H3 33B Omni-Transformer
- H3 in-context 2K regeneration
- H3 public weights/code
- M2.5 coding/productivity advances
- M2.7/M3 evolution
- Public MCP tooling
- MiniMax Agent workspace
- Cross-platform model integrations
Business
- Company established 2021/22
- Global consumer expansion
- 2024 reported $600M financing
- Hailuo creator growth
- VEED integration
- Envato integration
- Monks partnership
- Film-festival programs
- Hong Kong IPO process
- HKEX listing
- FY2025 revenue $79M
- 73% FY2025 revenue outside mainland China
- 236M+ users Dec 2025
- 214K enterprise customers/developers Dec 2025
- H1 2026 revenue $116.6M
- Open Platform/enterprise $73.9M H1 2026
- AI-native products $42.6M H1 2026
- 300M+ current users
- 2M+ current enterprise/developers
- H3 public release
37. Top 30 Verified Facts
| # | Fact | Evidence |
|---|---|---|
| 1 | MiniMax Group Inc. trades on HKEX as 00100. | VERIFIED |
| 2 | Trading began 9 January 2026. | VERIFIED |
| 3 | Shanghai is the PRC headquarters/principal place of business. | VERIFIED |
| 4 | Dr. Yan Junjie is founder, chairman, CEO and CTO. | VERIFIED |
| 5 | Yan worked at SenseTime for more than six years. | VERIFIED |
| 6 | Yan has a mathematics degree from Southeast University. | VERIFIED |
| 7 | Yan has an AI doctorate from CAS Institute of Automation. | VERIFIED |
| 8 | Yun Yeyi is an executive director and COO. | VERIFIED |
| 9 | Zhao Pengyu leads LLM research and engineering. | VERIFIED |
| 10 | Zhou Yucong leads visual-model research and engineering. | VERIFIED |
| 11 | Mission: Intelligence with Everyone. | VERIFIED |
| 12 | Values: No Shortcuts, User-in-the-Loop, Tech-Driven. | VERIFIED |
| 13 | Hailuo AI is an AI-native product in the MiniMax ecosystem. | VERIFIED |
| 14 | Hailuo 02 launched 18 June 2025. | VERIFIED |
| 15 | Hailuo 2.3 launched 28 October 2025. | VERIFIED |
| 16 | Hailuo 2.3 Fast targets faster/lower-cost batch creation. | VERIFIED |
| 17 | Media Agent evolved from Video Agent. | VERIFIED |
| 18 | Director models launched 3 March 2025. | VERIFIED |
| 19 | MiniMax H3 launched 31 July 2026. | VERIFIED |
| 20 | H3 generates native stereo audio with video. | VERIFIED |
| 21 | H3 supports 4–15 seconds in the open release. | VERIFIED |
| 22 | H3 runs at 24 FPS. | VERIFIED |
| 23 | H3-Omni-Transformer is a 33B dense Transformer. | VERIFIED |
| 24 | H3 uses Qwen3-VL-32B as its encoder according to MiniMax documentation. | VERIFIED |
| 25 | H3 weights/code are public under a custom community license. | VERIFIED |
| 26 | H3 general applicable territory excludes US/EU/UK/South Korea. | VERIFIED |
| 27 | H1 2026 revenue was $116.6M. | VERIFIED |
| 28 | H1 2026 Open Platform/enterprise revenue was $73.9M. | VERIFIED |
| 29 | FY2025 revenue was $79.0M. | VERIFIED |
| 30 | FY2025 revenue outside mainland China was about 73%. | VERIFIED |
38. Strategic Insights & Final Verdict
| Question | Evidence-based answer |
|---|---|
| Why adopt Hailuo? | Fast multimodal video creation, reference/control features, native audio, relatively low inference cost and increasingly integrated workflows. |
| Strongest customer-facing use case? | Commercial/creator video generation and iteration is the clearest public Hailuo use case. |
| Which industries show strongest public evidence? | Creative/advertising, film/media, e-commerce, software/creative SaaS and creator economy. |
| Strongest measurable benefit? | Speed/cost efficiency is the most consistently quantified product benefit; the 50% Fast-model claim is company-reported. |
| Highest ROI case? | Cannot be independently calculated because customer-level spend and complete cost baselines are not public. |
| Greatest scale? | MiniMax reports 300M+ individual users, 2M+ enterprise customers/developers and 600M+ cumulative video generations by end-2025; these are not Hailuo-only user counts. |
| Strongest localization? | Global distribution is clear, but quantified Hailuo-specific localization case data is limited. |
| Best fit? | High-volume short-form commercial/social/product creative teams, creators and developers needing API-based generation. |
| Who may not need it? | Teams needing long-form continuity, exact frame-level finishing or unrestricted H3 deployment in excluded territories may need additional tooling. |
| Biggest limitations? | Reliability/consistency, iteration cost, long-form continuity, rights/compliance, moderation and changing model access. |
39. Research Limitations & Fact-Checking Rules
- MiniMax-wide users/revenue are never treated as Hailuo-only metrics.
- Integrations, partners, demonstrations and creator collaborations are not automatically classified as customers.
- Benchmark scores are time-sensitive and do not equal customer success.
- Legacy API prices remain documented even as H3 becomes the flagship public model.
- Funding amounts are not summed where round-by-round primary evidence is incomplete.
- “Not publicly disclosed” is used instead of guessing missing employee counts, customer counts, training datasets or ROI.